eigenloss

11 ms·

91.
by eigenloss 8y ago
The current in question increased from 0 to 4 amps in approximately 100 femtoseconds (10^-13 seconds). When generating fields this large, the coil's destruction is guaranteed, so to create a large maximum field magnitude, you have to r
92.
by eigenloss 8y ago
> Part of the problem with blogs is that they're less rewarding than Facebook and Twitter: your post may perhaps get some thoughtful responses but it doesn't get immediate likes. To me, this is not a problem. People should not
93.
by eigenloss 8y ago
You're both talking about Tumblr like it's dead. Tumblr will never die.
94.
by eigenloss 8y ago
> build: failing :(
95.
by eigenloss 8y ago
This is the worst solution posed here. If we had no FCC, large companies would be thrown into an arms race to pump as much RF as possible into the air while smaller ones (radio stations, anybody?) would simply be drowned out and crushed.
96.
by eigenloss 8y ago
https://github.com/cliffordwolf/picorv32
97.
by eigenloss 8y ago
It looks like this webpage has been overwhelmed at this point.
98.
by eigenloss 8y ago
...except in Rust, frustratingly.
99.
by eigenloss 8y ago
We have quad-floats (128-bit) now which seem to work alright.
100.
by eigenloss 8y ago
A cube (square prism) enclosing a cubic millimeter has more surface area than a sphere enclosing the same volume.
101.
by eigenloss 8y ago
A relevant point here is that video compression almost always destroys the signal relied upon in this research.
102.
by eigenloss 8y ago
> The amount of capital required (both human and material) is too large to achieve colonizing Mars, even in a small capacity. Sending people to Mars would cost a few times as much as the construction (inflation adjusted) of the US inters
103.
by eigenloss 8y ago
You cannot prove the results of decades of physics and mathematical debate wrong with a few paragraphs of emotional arguments about dogma or some cheap mockery of a TV show.
104.
by eigenloss 8y ago
"More than one hundred years after he published his paper setting out the equations of general relativity, Einstein has been proved right once more — in a much more extreme laboratory than he could have possibly imagined!" This so
105.
by eigenloss 8y ago
Physical damage caused by user negligence and perceived defects in design or manufacture due to third-party repair service are two different things.
106.
by eigenloss 8y ago
For those confused, title was changed from something about "tiny robot Olympics."
107.
by eigenloss 8y ago
As far as hardware hacking goes, this is tremendously unimpressive work. They didn't even check for PCB antennas, which would have been trivial. It would probably take six months, minimum, of real work to actually airgap a Macbook (wit
108.
by eigenloss 8y ago
Thanks for the reply!
109.
Show HN: Source release for “Is this loss? A TFLite app to detect Loss.jpg”
7 points by eigenloss 8y ago|1 comments
110.
by eigenloss 8y ago
I've been looking into re-implementing my Loss.jpg detector in TensorFlow.js, but I'm a little concerned with the speed achievable in the Chrome sandbox. Loading the linked page on my laptop was tremendously slow and shot my CPU u
111.
by eigenloss 8y ago
Obviously, loss/not loss isn't a genuinely useful classification. Something to automatically help users identify forum-specific images (e.g. poisonous/harmless spiders, bedbugs/beetles, snakes, etc.) would be useful, and
112.
by eigenloss 8y ago
A very similar demo is available on the Tensorflow website, and the sources are available in the main TF repo. My sources will probably go public eventually. https://www.tensorflow.org/mobile/tflite/demo_android A
113.
by eigenloss 8y ago
Not a bad idea. The "not loss" training set actually contains a few hundred images scraped from there. There are plenty of subreddits and online forums where a single-purpose lightweight CNN like this one could be enormously produ
114.
by eigenloss 8y ago
Yes; I'll be doing one or both of the above shortly. There are tutorials online for most of the process, but the non-Google-maintained ones are generally very out of date. The app itself draws heavily from the TFLite Java Demo App, whi
115.
by eigenloss 8y ago
Framed properly, it does detect what you posted: https://imgur.com/a/0wdqPea . Positively identifying abstract geometric loss while rejecting non-loss text and drawings was something I had to explicitly optimize the tra
116.
by eigenloss 8y ago
Training was a bit of a process. The architecture I use now is an 8-bit quantized mobilenet_v1_050_224 retrained on about 1000 versions of Loss.jpg (some quite similar, but not exact duplicates) and about 10,000 other images. All the images
117.
by eigenloss 8y ago
Thanks lccarrasco! It does work surprisingly well for such a tiny CNN. Props to the MobileNet team at Google.
118.
by eigenloss 8y ago
Maybe. I don't see any reason not to! Edit: I can't reply to any other comments at the moment due to low karma. In response to wrinkl3 and others, I will likely be sharing more details about the development process and network in